Titanium matrix composite (TMC) rings are useful in high temperature rotating parts, such as turbine engines, where specific stiffness and strength are critical to design. While affordability issues generally have hampered the use in production of these materials, one TMC fabrication method has shown promise. According to this method titanium wire and silicon carbide (SiC) fiber are combined to form a hoop reinforcement array. Methods for fabricating TMC rings in this way have been described in U.S. Pat. No. 5,763,079 to Hanusiak et al. and U.S. Pat. No. 5,460,774 to Bachelet. These two patents describe different approaches to achieve the same end. However, both also restrict manufacturing flexibility in ways critical to design.
The method described by Hanusiak et al. is illustrated in FIGS. 1A-1C. In this approach, the combination of wire 3 and fiber 4 is restricted to a one-to-one ratio, but the wire diameter and the fiber diameter can be different as long as the wire 3 diameter is greater than that of the fiber 4. The selection of wire and fiber diameter establishes the fiber fraction in the resultant composite. For example, using a 0.007 inch diameter wire and a 0.0056 diameter fiber results in a composite with a fiber fraction of 30%. In accordance with Hanusiak et al., the assembly consists of one tape containing all wire elements and one tape containing all fiber elements combined to form two layers per ply. Each tape is made up of equal-sized elements, but the elements in the first tape do not have to be the same size as the elements in the second tape. The assembly is built up using alternate tapes of each type applied to a winding core in such a way that adjacent fibers 4 do not come in contact with each other. The advantage of the structure according to Hanusiak et al. is that the ratio of wire-to-fiber diameters can be varied such that composites with fiber fractions between 35% and 45% can be readily fabricated. Such a range of fiber fractions is particularly desirable for effective ring construction. The disadvantage of the structure according to Hanusiak et al., however, is that the assembled array contains about 20% void, which is particularly detrimental in thick parts because it allows for undesirable cusp formation during metal movement. Moreover, the structure according to Hanusiak et al. has been shown to be organizationally unstable during a consolidation cycle to remove the void content of the TMC part.
FIG. 1A shows a cross-section of a composite ring structure 1 according to Hanusiak et al. wherein there is maximum fiber spacing such that wires 3 touch in the height direction only. FIG. 1B shows an embodiment in accordance with Hanusiak et al. wherein there is median fiber spacing such that the fibers are spaced equally in width and in height. FIG. 1C depicts yet another configuration of a structure in accordance with Hanusiak et al. wherein there is minimum fiber spacing and wires 3 touch each other in the lateral or width direction only.
The method described by Bachelet is illustrated in FIGS. 2A-2C. According to Bachelet, the wire/fiber combination is restricted to a two-to-one or a three-to-one ratio. Additionally, in all of the examples disclosed by Bachelet, the wire diameters are limited to the same dimension as the fiber diameters. All assemblies utilize two layers per ply and fall into three types as shown in FIGS. 2A-2C.
Specifically, as shown in FIG. 2A, each layer is made up of fibers 4 separated by two equivalent-diameter wires 3, and the second layer is laterally indexed so that the fibers 4 nest between the two wires 3 in the layer below.
In other variations of the Bachelet structure, as shown in FIGS. 2B and 2C, one layer is made up of fibers 4 separated by one equivalent-diameter wire 3. The second layer is made up of all wires 3 of the same diameter as the fibers 4 in the first layer. The advantage to the Bachelet approach is that the void content is only about 10%, and the array apparently is organizationally stable during subsequent consolidation steps. Furthermore, the Bachelet approach, because of the resulting relatively low void fraction, may be desirable for thick parts since there is a lower tendency for cusp formation along the TMC perimeter. The disadvantage, however, of the Bachelet approach is the limitation in the examples to equal diameters for the wires and fibers, which limits the fiber fraction to 25% or 33%. These fiber fractions are not in the most desirable range from a design standpoint. That is, in many designs, a 40% fiber fraction is desirable to achieve a useful performance increase.
Additionally, all examples disclosed in the Hanusiak et al. and Bachelet patens are limited to equal-sized elements in any single layer. Although those references do not specifically exclude the case where elements in a layer may have different diameters, neither reference addresses the special problems associated with such a structure. Namely, when dissimilar-sized elements are provided in a single layer and all elements in a layer are applied to the winding core simultaneously there occurs an inherent stacking, or organizational, instability.
It is noted that simultaneous application of all elements in any single layer is a specific requirement of Bachelet. Bachelet apparently applies this constraint to control the element spacing in the first layer, since the reference fails to describe any other method for spatially controlling elements in the first layer on a winding mandrel. This also implies that the elements in the first layer are touching in order to effectively fulfill the positioning goal. Subsequent layer element positions are thus defined by gaps created between elements in the first layer. Given a first layer with touching elements, and dissimilar wire and fiber diameters, subsequent layer elements will typically lose their track due to nesting site ambiguity and the assembly will fall into disarray. FIGS. 3A-3C depict how a second layer of non-equal sized elements might be disposed on a first layer of non-equal sized elements and how, ultimately, after several layers have been applied, substantially all order is lost (FIG. 3C). That is, the non-equal element size in a given layer creates competition for nesting sites if the subsequent layer elements arrive at the same time.
Thus, there is a need for an improved method for achieving low void content in a stable array, concurrently with flexibility in fiber fraction between about 0% to 70% and preferably between about 30% and 45%.